Artificial Intelligence in Healthcare is attracting increasing attention: chatbots are often used as an alternative to traditional psychotherapy. However, mental health experts warn that such use poses significant risks to users, especially those vulnerable to mental disorders.
This is reported by Business • Media
Real Risks: Cases from Belgium and the USA
The first alarming signals came from high-profile incidents in various countries. In 2023, in Belgium, a man suffering from “eco-anxiety” took his own life after prolonged interaction with a chatbot. In another case in Florida, USA, a 35-year-old man died during a confrontation with police, believing that artificial intelligence had embodied a fictional persona.
Research from Stanford University has shown that large language models (LLMs) often give dangerous advice to people with delusional or suicidal thoughts. During experiments, chatbots not only responded to inquiries about suicide but could also recommend specific locations for it.
Expert Opinion: Chatbots Will Not Replace Professional Help
Experts from the UK emphasize that artificial intelligence algorithms are not only prone to giving incorrect answers but can also exacerbate psychotic manifestations as they attempt to reinforce the user’s own beliefs.
Sahra O’Doherty, president of the Australian Psychological Association, notes that many people turn to ChatGPT as an additional support tool. At the same time, due to financial difficulties, chatbots often completely replace traditional therapy.
“Artificial intelligence merely reflects what you put into it and can drive a person deeper into the rabbit hole of their own fears,” the expert emphasized.
Some researchers acknowledge that chatbots can be useful as motivational coaches but warn of long-term threats. Dr. Raphael Miller highlights that becoming accustomed to a “flattering and compliant conversational partner” affects the development of real communication skills in the younger generation.
Experts agree: despite their accessibility and apparent support, chatbots cannot replace qualified psychiatric help. They advise combining the use of artificial intelligence with critical thinking and seeking professional assistance when needed.
Previously, the Future of Life Institute (FLI) concluded that developers of AI models are “absolutely unprepared” for the potential threats posed by superintelligence.